Reducing uncertainty in unconventional reservoir hydraulic fracture modeling: A case study in Saudi Arabia

Saudi Aramco is exploring and appraising liquid rich and dry gas unconventional reservoirs. Part of this effort has focused on understanding the effective fracture geometry and its role in optimal well spacing and reservoir drainage. A review of collected data and existing hypotheses indicated that there are multiple models for optimal production yet each calls for a dramatically different approach. Long term flow data is recommended in such cases but this is often complicated by the fact that decisions in a fast paced environment might rely on short production history. A field pilot in a carbonate-rich ultra-low permeability source rock in Saudi Arabia incorporated multiple hydraulic fracture monitoring techniques to validate outputs from a classical linear elastic fracture model. All employed monitoring techniques had their results assessed in conjunction with their strengths and weaknesses for more inclusive interpretations. From the reservoir flow regimes and the various hypotheses on effective fracture geometries, the extent to which non-uniqueness impacts determination of the most probable fracture geometry and matrix permeability was investigated. Long and short hydraulic fractures were used to match the short production history with reservoir simulation models under different sets of assumptions. The contradictions in the assumptions and difficulty in transferring fracture geometries, calibrated by the monitoring results, to production data modeling, were used to gauge the validity of associated effective fracture geometry hypotheses. From the pilot, microseismic monitoring indicated that more stimulation volumes yielded longer microseismic geometries. Offset well stimulation induced pressure interference in adjacent wells. Cross-well hydraulic communication illustrated by chemical and oil tracers corroborated the longer fractures mapped by microseismic events and also predicted by linear elastic fracture models. In addition, the sequential opening of the pilot pad wells induced subtle production-induced pressure interference in the early flow period. While the subtle production-induced pressure interference was initially thought to be the ground truth for adequate well spacing, it was shown that it was not unique due to the fact that ultra-low matrix permeability can prolong onset of fracture interference. The hydraulic fracture monitoring methods showed identical results for the estimated hydraulic fracture half-length which allowed the constraining of the parameter. Reservoir simulation was used to match the short production history with both short and long planar fractures even though the former had not been validated by most of the observed data. Consideration of all observed data, dimensionless fracture conductivity for stimulation design criteria and the reservoir simulation assumptions provided validation for long effective fractures. This study shows the contribution of a diverse surveillance data acquisition in reducing uncertainty in unconventional reservoir fracture modeling for realistic reservoir simulation and development solutions. In addition, insights derived from this work provided important clues for testing the validity of hypotheses in unconventional reservoir studies.